Cthesis/food-delivery-operations-analysis

Food delivery operations analysis: 10,000 orders across 5 South African cities — data cleaning, EDA, statistical findings, and business recommendations. # Food Delivery Operations Analysis A food delivery platform operating across five South African cities (Johannesburg, Sandton, Midrand, Soweto, Pretoria) had 10,000 raw order records and no reliable answer to two questions: what actually makes a delivery slow, and why does roughly one in three orders cancel outright. **Headline finding:** delivery time has no measurable relationship with either delivery distance (r = -0.003) or restaurant prep time (r = -0.001) — ruling out the two most common explanations. Separately, the 33.5% cancellation rate is nearly identical across every city and payment method, pointing to a systemic cause rather than a local one. Full write-up with charts: case study on the portfolio site. ## Project Map | Folder | Contents | |---|---| | `01 Documentation/` | README, Data Dictionary, Methodology, Cleaning Log | | `02 Data/` | Raw and cleaned CSVs | | `03 SQL/` | Schema, analysis queries, SQLite database | | `04 Python/` | Cleaning pipeline (`clean_data.py`) and EDA script (`eda.py`) | | `05 Dashboard/` | Dashboard status notes | | `06 Reports/` | Executive Summary, Business Report, Technical Report, Findings, Recommendations, Future Improvements (editable source docs) | | `07 Presentation/` | Stakeholder deck and speaker notes | | `08 Images/` | All 15 EDA charts, generated by `eda.py` | | `09 Assets/` | Computed metrics (`metrics.json`) and report-generation scripts | | `10 Final Deliverables/` | PDF exports of the four core documents — start here for the client-ready version | ## Dataset - **Source:** `food_delivery_dataset.csv`, 10,000 raw rows → 9,793 after cleaning - **Columns:** 15 raw → 21 cleaned (6 derived) - **Date range:** 1 January 2026 – 1 May 2026 (4 months) - **Currency:** South African Rand (ZAR) - **Tooling:** Python, pandas, matplotlib, SQLite ## Reproducing the Results ``` 04 Python/clean_data.py # full cleaning pipeline, prints a step-by-step log 04 Python/eda.py # generates all …